November 2, 2026
300 places in 118 of 485 files
Tools that quietly ignore a misnamed argument give quiet wrong answers.
The tool now reports the keys it ignored. A scanner found the same shape in 300 places across 118 of 485 files. The test runs both ways: the old version is flagged, the fixed one isn't.
Every tool an AI calls should say what it didn't use. Silence looks like success.
Source: Commit e5e5098c.
October 8, 2026 Didn't work
0 trained versions that passed their grade
We trained a small local copy of our assistant for ten weeks. It never passed.
In ten weeks, no trained version passed its grade, and nothing in the system relied on it. We retired it on October 8. The files are kept, so it can come back.
A model that can't beat its own test isn't a feature yet, however good the idea sounds. We kept the test and dropped the model.
Source: Retirement record, October 8, 2026.
October 29, 2026
868 pieces proofread in under three hours
An agent with nothing to do repeats itself. Work fixed it; filters didn't.
Given live websites to proofread, she read all 868 pieces in her queue in under three hours and reported what she found.
An agent needs material, not just freedom. When it loops, give it a job before you add a rule.
Source: Commit 24c985ed, October 29, 2026.
October 14, 2026 Didn't work
93 forecast rows from one template, all at 60%
A confidence number about our assistant came from a template, not from the assistant.
All 93 rows came from one template line with a fixed 60%. The assistant never stated them. Its own forecasts since October 2: 10 right out of 18, at 68% stated on average.
Eighteen forecasts is too few to say whether it runs over- or under-confident. We struck the old number and say "we don't know yet".
Source: Forecast audit, October 14, 2026.
